DE60012739T2 - SPECTRAL MODEL FOR PHOTOGRAPHIC PRINTING BASED ON A COLOR CONCENTRATION - Google Patents

Description

The The present invention relates to modeling a photo printing process. and more specifically, modeling the process and device for producing a photo print based on a photo color negative.

BACKGROUND THE INVENTION

At the Photographic printing produces a photo print from a photo negative. at The photo printing process has a different effect when producing the photo print Factors together, including Properties of exposure light bulb and process, of Printing paper and the colors. To a photo printing by means of a photo printing model To accurately characterize each of these real aspects of the printing process be modeled.

To the Scanning a photographic color negative and generating corresponding ones Output digital values can be used by a scanner. scanner generate output digital values to the spectral transmission factor an input pattern. A conventional one Scanner illuminates a permeable Target using a light source. The scanner integrated Light passing through the pattern and through a set of spectral selection filters penetrates. The integrated products can with the help of electronic Device and software to produce output digital values. These Output digital values can for the characterization of photo media with print and color models be combined. conventional However, photo characterizations are based on density and are not channel independent. A channel independent Characterization would create a more flexible model.

Further can different reflection spectra of the eye as the same color be perceived. Similar can different spectra generate the same digital values. This Effect is called "metamerism". metameric Patterns can produce different colorimetric sensitivity values, when viewed under different viewing conditions or sampled. Consequently, when modeling scanners many different reflection spectra generate the same RGB values. For a more accurate estimate of a reflection spectrum, it is desirable to set the parameters reduce to metameric matches to avoid.

US-5,790,280-A describes a digital image recording method in which Image data of an original based on an image tag value processed, such as based on an average density of a Object area of the original, derived from the image data becomes. A basic mapping record set for each Pixel is used based on the image tag value determined. Based on the basic image record amount of a value of the image data assigned to a pixel Recording data of this pixel is determined, and the recording data be in an image control amount with reference to a Converted table. The image control amount is used in an image recorder Record the reproduction image used.

US-A-5,521,723 describes a color image reproduction system with a color printer that can make a color print that looks like the original illustration recorded with the help of a photo process. This effect will achieved by means of a transformation, in a three-dimensional Table embodied which was formed as follows: a) a transform for transforming the color digital image in original scene exposures; legs Transform to transform the video scene exposures into Photo scene exposures; c) transforming to transform of photo scene exposures in channel independent densities; d) a transformation to Transform channel independent Densities in reproduced densities; and e) a transformation to Transforming the reproduced densities into control signals for control the color hardcopy output device.

SUMMARY

The The invention provides a method and a computer program according to the claims 1 and 24, which is a technique for predicting a reflection spectrum to implement a photo print. The dependent claims relate on individual embodiments of the present invention.

The Technology according to the present Invention models the pressure based on data from a source image, such as a color negative or color positive.

The technique is further adapted to model output imaging on transmissive media. The data is preferably generated by means of an imaging device, such as a scanner or digital camera. The technique uses color concentration, creating a channel independent basis for the spectral model. When the technique is used in conjunction with a color management system, the technique allows for previewing or manipulating the photographic images or experimenting with them Photo processes using simulation. The user can control the simulated processing by means of a table and, optionally, by means of a print illuminant color temperature which controls the exposure time and the spectral content of the print illuminant. The user may also control other aspects of the simulated process by modifying parameters such as the color filters, the illuminant spectrum and the spectral sensitivities or absorption spectra of the paper.

The preferred spectral model comprises three term sections: scanners and movie models, a photo printer model, and a photo paper model. The scanner and Film section digitally converts data from one sample Color negative on a calibrated scanner in predictions of spectral transmission factor for each Pixel. In alternative implementations, the scanner model becomes by alternative models of image capture devices replaced. The photo printer section uses a table and optionally a color temperature and calculates the spectrum of exposure as a function currently. The photo paper section converts the spectral energy of the illuminated negative first in color concentration estimates and subsequently into a predicted reflection spectrum for the simulated pressure.

Generally In one aspect, the technique comprises the following steps: Convert of digital values from scanning a photographic negative into a film transmission spectrum using a photographic negative film model that is a medium of the photographic negative corresponds; Estimate a light source illuminant spectrum over time using an exposure table; Estimate of paper spectral sensitivities of a photo paper that accompanies the photo print corresponds; Integrating spectral products of the film transmission spectrum, the exposure illuminant spectrum and paper spectral sensitivities over the Time; Convert the integrated spectral products to logarithmic integrated products; Convert the logarithmically integrated Exposures in color concentrations; and converting the color concentrations into a predicted range of reflection.

The Implementations according to the present Invention provide one or more of the following advantages: the calculation based on the color concentration is simple and directly; the predicted spectrum is an accurate model of the photo media and the development process; the model allows realistic experimentation, a preview and a global or selective manipulation of Photographs without physically exposing a negative and then one Develop photo printing from the negative; experimenting with new algorithms for an automatic color correction and the investigation of material properties or printer features are simplified; and the provision complex parameter is improved by using targets which are easy to create and measure. At given exact Parameters in the model can furthermore physical constraints attenuated the photo printing process and better mappings than those generated by analogue Print produced. For example, adjusting the Exposure area compression blocked shadows and overly bright Avoid digits, and exposures that ignore side effects can produce more powerful images. Furthermore, the simulated Processing can be implemented as a computer program to a virtual Photo printer device including software controls to create that mimic the behavior of an actual device.

following The invention will be described in more detail with reference to the drawing. Show it:

1 a block diagram of an inverse model according to the invention;

2 a flow chart of the prediction of a reflection spectrum according to an embodiment of the invention;

3 a block diagram of scanner and movie models;

4 a flowchart of a forward sampling model;

5B a flowchart of an alternative implementation of a forward sampling model;

6 a flowchart of estimated parameters for the inverse model;

7 a graphical representation of the spectral reflectance of a target with a neutral color wedge;

8A a flowchart of estimated parameters for the inverse model;

8B a flowchart of estimated parameters for the inverse model;

9 a block diagram of a photo printer model;

10 a flow chart of the photo printer model;

11 an example of an exposure table and

12 a block diagram of a photo paper model.

Same Reference numerals and designations in the various drawings denote the same elements.

PRECISE DESCRIPTION

In one embodiment of the invention, a computer system predicts the reflection spectrum of a photographic print based on a photographic negative. As in 1 is shown, the computer system comprises a spectral model 100 of a photo printing process, scanning and film models 105 , a photo printer model 110 and a photo paper model 115 , The models 105 . 110 . 115 can be implemented as logically or physically separate components or as a single component in which the models are only abstractly distinguished. Thus, the mapping of calculations and equations described below is an example implementation of a variety of possible implementations.

The sample and film models receive target digital values that are generated by sampling the photographic negative (ie, obtaining and processing imaging information based on the spectral energy received from the exposed pattern) using a scanner (not shown) and convert them Values on a per-pixel basis in a film transmission spectrum. Alternatively, the scanning model uses 105 a color table for determining the target digital values. In an alternative implementation, the scan model is 105 is replaced by another model of an imaging device, such as a digital camera model or other scanner model. Examples of other spectral scanning models include those described by RS Berns and MJ Shyu in "Colorimetric Characterization of a Destop Drum Scammer using a Spectral Model," Journal of Electronic Imaging, Vol. 4, No. 4, October 1995, pages 360-72; by Dunne et al., in U.S. 5,149,960, Sept. 1992; CE Mancil, in "Digital Color Image Restoration", PhD. thesis, Univ. of S. Calif., Aug. 1975; and by Sherman, in U.S. 5,543,940, Aug. 1996. In a further alternative implementation, a transmission spectrum corresponding to a color negative can be achieved by direct measurement, such as with a sampling pass spectrophotometer.

The photo print model 110 models the spectrum of a light source over time. The photo print model 110 estimates an exposure illumination spectrum over the time period according to an exposure table. The exposure table may be supplied by a user or automatically determined, such as using statistics of the scanned negative data. The photo paper model 115 models the color absorption and reflection of a particular photo paper. The photo paper model 115 converts the film transmission spectrum, the exposure illumination spectrum, and the paper spectral sensitivities corresponding to the photographic paper in color concentrations. The photo paper model 115 further converts the color concentrations into a predicted reflection spectrum. The predicted reflection spectrum represents a reflection spectrum of a photographic print that would be achieved if the scanned photographic negative were to be scanned using the spectral model 100 modeled photo printing device and the modeled paper would be developed. Each of these models is described in more detail below.

As in 2 is shown, the computer system converts in a method 200 for predicting the reflection spectrum of the photographic printing, digital values of scanning the photo negative into a film transmission spectrum (step 205 ). The photo negative is preferably a color negative, but it may alternatively be a grayscale negative (eg, black and white) or a color or grayscale positive. To convert the digital values, the computer system uses a photo negative film model corresponding to the photo negative media. The computer system uses a user supplied exposure table and a model of a photo printer to estimate an exposure illumination spectrum over the time period (step 210 ). The computer system may also optionally use a color temperature to estimate the exposure illumination spectrum. The computer system estimates paper spectral sensitivities according to a photo paper (step 215 ). Alternatively, the paper spectral sensitivities may also be supplied or predetermined. The computer system integrates spectral products of the film transmission spectrum, the exposure illumination spectrum and the paper spectral sensitivities (step 220 ). The computer system converts the integrated spectral products into logarithmic integrated exposures (step 225 ) and convert the logarithmic integrated exposures to color concentrations (step 230 ). To complete the process, the computer system converts the color concentrations into a predicted reflection spectrum (step 235 ).

Scanner and film model

In one implementation of the scanning and photographic negative models, a computer system models the colorimetric behavior of a graphics scanner when scanning a pattern, the photon gativs. As in 3 is shown, the computer system uses an inverse model 300 which converts target digital values from scanning a photographic color negative into an estimated spectrum. The scanner includes at least one output channel (eg, R, G, and B) and generates target digital values for each channel. The estimated spectrum may be an estimate of the spectral reflectance or spectral transmission factor, depending on the application. The inverse model 300 includes a forward model 305 which estimates digital values for each output channel of the scanner based on a provided spectrum. The inverse model further includes a media model tailored to the medium of the pattern 310 , which transforms media coordinates into an estimated spectrum. The media model may be a model for reflective or transmissive media. For a photographic color negative, the media model is a photographic color negative film model corresponding to the specific medium of the photographic negative. The media model 310 receives color negative color absorption spectra to model the color absorption of the photographic negative. The negative color absorption spectra may be provided by the film manufacturer or estimated as described above. The inverse model 300 repeatedly estimates coordinates in a colorant space corresponding to the photo negative using a search engine 315 is used. The search engine 315 searches in the colorant space, based on an error value 320 between target digital values from the scanner and estimated digital values from the forward model 305 , The inverse model 300 converted using the media model 310 estimated media coordinates into an estimated spectrum and transmit the estimated spectrum to the forward model 305 , The forward model 305 converts the received spectrum into estimated digital values. The inverse model 300 compares the estimated digital values to the target digital values by an error value 320 to investigate. If a stop criterion or stop criteria has not been determined, the search engine searches 315 new media coordinates based on the error value 320 and the process begins again. The stop criterion may be a specified tolerance for the error value 320 or, alternatively, to another criterion, alternatively to another criterion, such as a number of iterations. If the stop criterion has been met, the inverse model evaluates 300 the spectrum as the spectrum that sufficiently corresponds to the transmission spectrum of the pattern.

As in 4 is shown, estimates a computer system in a method 400 in which the spectral reflectance component of a color negative scanned by a scanner is modeled by identifying a predicted spectrum, media coordinates within a colorant space of a media model corresponding to the color negative (step 405 ). The computer system estimates the media coordinates based on target digital values generated by the scanner. The estimated media coordinates are estimated to correspond to a spectrum that produces the target digital values. For example, the media coordinates of a conventional color negative film represent the concentrations of cyan, magenta and yellow hues. The estimation of the initial media coordinates is not critical. The inverse model should approach an acceptable resolution even with "bad" initial estimates. However, the probability of approaching an incorrect metameric match is reduced if the initial estimates are close to their final values. In a process of estimating initial media coordinates, constant initial values from the middle of valid ranges are used for the media coordinates. In an alternative process, the target RGB values from the scanner or their inverted values are used as initial estimates. For example, in one implementation, the media model represents media with cyan, magenta, and yellow hues, and all media coordinates and target device values are normalized to a range of 0 to 1. In this case, the media coordinates can be in the form of inverted target RGB values C 0 = 1 - R tgt M 0 = 1 - G tgt Y 0 = 1 - B tgt where C ₀ , M ₀ and Y ₀ are the estimated initial _{media coordinates} and R _tgt , G _tgt and B _tgt are the target RGB values. Other processes known to those skilled in the art for estimating the initial media coordinates may also be used.

The complete system converts the estimated media coordinates into an estimated spectrum using a spectral media model (step 410 ). As previously described, the media model is tuned to the medium of the photographic negative. The computer system selects media coordinates that reside in the colorant space of the media model. Thus, the media model limits the corresponding spectra, and its use thus limits the resulting metamerism (ie, different spectral values that appear to produce the same color in a particular viewer or device) by estimating the media coordinates. The media model reduces the metamerism problem by reducing the coordination are forced into spectra that can not be observed on the medium, ignoring metameric spectra that the medium can not produce. Furthermore, the colorant space of the medium has less dimensions than a spectral measurement, making the search process more efficient over the colorant space of the medium. For example, in a CMY colorant room, the colorant room has three dimensions. A measured spectrum usually includes a large number of spectral wavelengths or wavebands, such as 16, 31 or 36.

For the color negative film, the media model uses the Beer-Bouguer theory to describe the relationship between the color concentration and the spectral transmittance factor. The Beer-Bouguer theory is described by Berns. This model can be expressed as follows: T n (λ) = T G (λ) exp [- {cK c (λ) + mK m (λ) + yK y (Λ)}] where c, m and y are the color concentrations, λ represents the wavelength, K _c (λ), K _m (λ) and K _y (λ) are normalized spectral color absorptivities, T _g (λ) is the spectral transmission factor of the substrate and T _n (λ) is the spectral transmission factor of a color correction. Unexposed film, due to the presence of capping couplers, can not represent either the true substrate or the maximum transmission factor in a given spectral band. Masking (ie, film making defects which, for example, provide a strong red exposure to create a cyan color concentration lower than that of the unexposed support) and other effects may result in certain colors being transmitted in some spectral bands and thus exceed that of the unexposed film. However, when evaluating film models derived from either unexposed film or the maximum transmission factor as a base, there is a slight difference in their ability to reproduce measured color negative transmission spectra. Therefore, the unexposed film transmission spectrum is preferred as the substrate transmission permeability factor since the unexposed film transmission spectrum is gentler.

at non-digitally addressed media, including color negative film generating single color exposures to determine each color absorption spectrum often difficult. Statistical methods, such as the factor study, are preferred to determine the color absorption spectra. E. R. Malinowski and D.G. Howery describe a method of factor examination in "Factor Analysis in Chemistry, John Wiley & Sons, New York, 1980, chapter 3. The application of factor modeling for media modeling includes a principal component study, which is a set of Media color patterns are applied to the absorption spectra, followed by a rotation of the main directions. The principal component investigation provides directions that define a space of lesser dimension, which in turn describes the pattern data deviation. By the rotation The main components should distribute the deviation evenly among the components be what often leads to factors of physical importance, such as individual ones Color absorption spectra.

Preferably, factoring is applied to absorption spectra of color negative patterns that span the gamut of a film recorder. The absorption spectrum K ⁱ _mix (λ) of the i-th pattern is calculated from the measured transmission spectrum T ⁱ _n (λ) and can be expressed as follows: K i mix (λ) = - 1n [T i n (Λ) / T i G (Λ)]

The Principal component investigation generates one eigenvector for each dye in the movie. For a color negative film thus produces the principal component study three eigenvectors, one each for the cyan, magenta and yellow hue. These vectors are described with the help of those described in Malinowski Target rotation transformed. Normalized absorption spectra of step wedges are for uses each channel of the film recorder as the target spectra (i.e. the target vectors for the target rotation). To modify the target absorption spectra a manual setting can be used. Such Setting can be used to compensate for the fact that the step wedge patterns are not single color exposures. The Rotated vectors may preferably essentially reproduce the entire absorption data set. To improve this reproduction, the vectors may have negative concentrations require (often in the order of magnitude of 5% or less of the maximum concentration). The rotated Spectra can Also include negative absorption areas to the training set to reproduce the pattern. By enforcing the concentrations the film models, the bipolar nature of the concentration is unproblematic.

After the computer system has received the estimated spectrum from the media model, it converts the estimated spectrum into estimated digital values using a forward model of a scanner (step 415 ). The computer system transmits the estimated spectrum to the forward model, and the forward model predicts digital values that the scanner would produce for the transmitted spectrum in the device-dependent coordinate space of the scanner. The forward model includes filter functions and channel-independent sound reproduction curves ("TRCs") specific to the scanner. The forward model optionally includes correction functions to represent channel interactions not considered in other sections of the forward model. The forward model will be described in more detail below.

The computer system calculates an error value between the target digital values and the estimated digital values (step 420 ). The computer system preferably calculates the Euclidean distance between the two sets of digital values. The error value may be a collection error value, including the digital values for all channels, or one error value for each digital value. Alternatively, the calculation of the error value may vary for each digital value. If a specific stop criterion is met, the estimated spectrum that generated the current estimated digital values is output as the predicted spectrum (step 425 ). The stop criterion may be fault tolerance. In an alternative implementation, each digital value has a different tolerance. In such an implementation, the estimated spectrum is the predicted spectrum if each error is less than the corresponding tolerance. If the stop criterion is not met, the computer system searches the colorant space of the media model by updating the current media coordinates based on the error to determine new media coordinates (step 430 ). The search engine preferably maintains statistics about the estimated media coordinates and errors to establish a "fault surface" model. The search engine uses the accumulated error surface model and the current error to determine new estimated media coordinates. The search engine preferably uses a numerical minimization technique such as the Powell method or the BFGS method described by WH Press et al. in "Numerical Recipes in C", second edition, Cambridge University Press, 1992. This technique is further limited so that only media coordinates within boundaries are identified according to the media model. As a result, the media model limits the search process and reduces metamerism. The computer system sends the new media coordinates through the media model (step 410 ) and repeat these steps until the stop criterion is met.

As in 5A The computer system receives in a forward model 500 a pattern spectrum, such as the estimated spectrum provided by the inverse model. The pattern spectrum is calculated using filter functions 510 multiplied 505 that correspond to each channel of the scanner. In an implementation where the scanner has red, green, and blue channels, the forward model includes three filtering functions, as in FIG 5A is shown. Each filter function 510 represents the product of a filter transmittance spectrum and a light source spectrum. The filter transmittance spectrum and the light spectrum need not be determined separately. Alternatively, the illuminant spectrum can be generated or measured directly, as described below. The filter functions 510 are estimated by training patterns as described below. The products of each filter function and the pattern spectrum are integrated over at least one wavelength, such as the visible wavelength, preferably a wavelength of about 380 to 730 nanometers ( 515 ). The integration results can thus be expressed as follows: r = ∫S (λ) R (λ) I (λ) dλ g = ∫S (λ) G (λ) I (λ) dλ b = ∫S (λ) B (λ) I (λ) dλ where r, g and b are the integration results; S is the pattern spectrum; R, G, and B are the filter transmittance spectra; I is the illuminant spectrum and λ is the wavelength. As mentioned before, R, G, B and I can be expressed as a filter function: F R (λ) = R (λ) I (λ) F G (λ) = R (λ) I (λ) F B (λ) = R (λ) I (λ)

The forward model 500 converts the integration results to digital values using tone reproduction curves ("TRC") 520 , The TRCs 520 are one-dimensional transformation functions, such as gain offset gamma functions, and can be expressed as follows. d r = T R [r] = [k gain.r r + k offset.r ] γr d G = T G [g] = [k gain.g g + k offset.g ] γg d b = T B [b] = [k gain.b b + k offset.b ] γb

The forward model 500 Optionally, these digital values from the TRCs 520 using correction features 525 as previously described.

5B shows an implementation of a forward model 550 in which the luminous spectrum is known or measured directly. The operation of the forward model 550 is essentially the same as the forward model 500 , this in 5A is shown and described above. The received pattern spectrum is multiplied by the transmitted luminous spectrum 552 , This product comes with filter permeability spectra 560 multiplied 555 that correspond to each scanner channel. In an implementation where the scanner has red, green and blue channels, the forward model includes three filter transmission spectra, as in FIG 5B is shown. The filter transmission spectra 560 are estimated from training patterns using the known light source spectrum. Thus, unlike the in 5A shown forward model 500 in which the filter functions representing the products of the filter transmittance spectra and a light source spectrum are estimated, the filter transmittance spectra 560 directly appreciated. As described above, the resulting products in step 565 integrated over the visible wavelengths of preferably about 380 to 730 nanometers and the integration results in digital value TRCs 570 converted. The forward model 550 Optionally, these digital values from the TRCs 570 as described above using correction functions 575 to adjust.

Before the computer system uses the inverse model to predict the spectral reflectance of the pattern, the computer system estimates the parameters of functions within the forward model. As in 6 shown is in a process 600 to estimate parameters, a target is scanned with a neutral step wedge to produce flat pattern values (step 605 ). The neutral stage sub-target preferably includes reflectance spectra that are substantially flat and a density range that includes the density range of most images that are to be digitized by the scanner. The spectra of the target are also measured by means of a spectrophotometer to generate flat pattern spectra (step 607 ). Depending on the nature of the medium, the spectrophotometer is either reflective or transmissive.

An example of reflection spectra 700 of a preferred flat spectra target is in 7 shown. The flat dashed lines 705 are mean spectral values for the measured values 710 for each range of visible wavelengths (about 400 to 700 nm).

For the i-th pattern in the flat spectra target, the correspondingly measured reflection spectrum S _i is approximated by a constant c _i , which is calculated by forming the mean value of the reflection spectrum S _i over the visible wavelengths. Thus, the above integrals can be described as follows (r being an example): r i = c i ∫ f R (Λ) dλ

The integral in this equation is also a constant, Φ, which is independent of the pattern spectrum S _i . Similarly, the TRC for r can be described as follows: d i r = T R [r i ] = T R [C i Φ] = [k gain.r c i Φ - k offset.r ] γr

By defining a new gain parameter k = _gain.r = k _gain.r Φ holds d i r = T ' R [c i ] = [k ' gain.r c i + k offset.r ] γr

Since S _i is substantially flat, this equation does not depend on the wavelength so that the TRC function can be sampled at the constant c _i . The samples thus generated are preferably used to estimate the parameters of the one-dimensional transformation functions with a constrained minimization technique using the flat pattern values (step 610 ). This estimate can be an iterative process that is repeated until a specific criterion is met. When a one-dimensional transform function is a gain-offset gamma function, the parameters are gain, offset, and gamma. The gain parameter may be too high by a factor of Φ, but this discrepancy can be compensated for by using an implicit scale factor introduced in estimating filter functions into parameters. The TRC parameters are preferably estimated on a channel independent basis.

To estimate the filter parameters of the filter functions, a second multicolor target having color patterns that span the scale of a particular printer device is scanned to produce multicolor pattern values (step 615 ). Further, the reflection spectra of the multicolor target are measured to produce multicolor pattern spectra (step 620 ). An approximation to the filter functions is preferably made in the forward model using Gaussian functions with the three following parameters: mean, standard deviation and amplitude. Alternatively, other functions may be used, such as cubic spline functions, blobs, summed Gaussian functions, delta functions, or a combination of these functions. For example, a combination of delta functions and Gaussian functions is preferred for a fluorescent lamp. A scan in eight Steps of RGB colorant space (512 samples total) converted to a CMYK color center space with parabolic gray component replacement can produce sufficient accuracy. Given predetermined estimated TRCs, as previously described, a limited minimization technique is preferably used to estimate the filter parameters (step 625 ). The estimated filter parameters minimize the mean color difference between values produced by scanning the multicolor target and values predicted using the estimated filter functions and measured spectra of the multicolor target. The computer system calculates filter digital values based on the measured spectra of the multicolor target using the forward model and the estimated TRCs and filter functions (step 630 ). In this estimation, the correction functions are preferably not used. The system calculates a filter difference by comparing the digital values with the multicolor samples from scanning the multicolor target (step 635 ). The filter difference is an average color difference, which is preferably calculated by determining the mean Euclidean distance between pairs of values. If a filter stop criterion is met, ie if, for example, the filter difference is within a specified filter tolerance (step 640 ), the estimation of the filter parameters is completed (step 645 ). If no filter stop criterion is met, ie if, for example, the filter difference is outside the filter tolerance (step 640 ), the computer system again estimates the filter functions (step 650 ) and repeat the previously described steps 630 to 640 , As with the TRCs, the filter parameters are preferably estimated on a channel independent basis.

In an alternative implementation, the TRCs and filter function parameters are repeatedly estimated in a design process until a convergence criterion is met. This convergence criterion may be met if a design difference is within a design tolerance, or alternatively after a certain number of iterations. As in 8A is shown, the TRC parameters are in an iterative parameter estimation process 800 as previously described (step 805 ). The TRC functions can alternatively be initialized as linear functions. Filter parameters are estimated as previously described, including sampling and measuring spectra of a multicolor target (step 810 ). The filter functions may alternatively be initialized as conventional R, G, and B filter functions, such as broadband Gauss functions centered at 450, 550, and 650 nanometers. The computer system estimates engineering digital values with the forward model using the estimated TRC functions and filter functions based on measured spectra of the multicolor target (step 815 ). Subsequently, a design difference between these design digital values and the digital values from sampling the multicolor target is calculated (step 820 ). If the convergence criterion is met, that is, for example, the design difference is within the design tolerance, the computer system is ready to model the scanner with an actual pattern (step 825 ). If no convergence criterion is met, eg, if the design difference is outside the design tolerance, the computer system again estimates the TRC and filter parameters. The estimated filter parameters are used to re-estimate the TRC parameters (step 830 ). Also, the filter functions are re-estimated using the re-estimated TRC functions as described above (step 835 ). The newly estimated TRC and filter functions are used to calculate new design digital values (step 815 ). The process is repeated until the design difference is within the design tolerance. Alternatively, the computer system also uses correction equations, such as stepwise multiple linear regression functions, to further refine the forecasts of the forward model.

In another alternative implementation, where TRC and filter function parameters are repeatedly estimated in a design process until a convergence criterion is met, the TRC parameters and filter function parameters are determined from training targets measured using a spectrophotometer. The convergence criterion may be met if a design difference is within a design tolerance, or alternatively once a certain number of iterations have been performed. As in 8B are shown in an iterative parameter estimation process 850 TRC and filter function training targets such as the flat and multicolor targets described above, sampled and measured by a spectrophotometer (step 855 ). The computer system determines (constant) averages for the TRC training spectra (step 860 ). The computer system estimates the TRC parameters using the constant values for the TRC training spectra (step 865 ). The computer system preferably estimates the TRC parameters using a constrained minimization technique. The computer system determines the filter parameters using the measured training patterns and the limited minimization (step 870 ). The computer system replaces the constant TRC training spectra with the actual spectral measurements and again estimates the TRC parameters (Step 875 ). The computer system uses the newly estimated TRC parameters to re-estimate the filter parameters (step 880 ). The computer system estimates design digital values using the forward model using the estimated TRC functions and filter functions based on the measured spectra of the multicolor target (step 885 ). A difference between these design digital values and the digital values from the scanning of the target is calculated (step 890 ). If the convergence criterion is met, eg, if the design difference is within design tolerance, the computer system is ready to model the scanner with an actual pattern (step 895 ). If this is not the case, for example, if the design difference is outside the design tolerance, the computer system calculates new TRC and filter parameters in the steps 875 and 880 , and the process is repeated until the design difference is within the design tolerance. Alternatively, the computer system also uses correction equations, such as stepwise multiple linear regression functions, which predict to further refine the forward model.

wobei d_r' der korrigierte Modellausgangswert ist; d_r, d_g und d_b die TRC-Ausgangswerte sind; f_i(d_r, d_g, d_b) die i-te Vorhersagefunktion ist; ϑ_i der i-te Regressionsparameter ist Und N_pred die Anzahl von Vorhersagefunktionen in der Regression ist. Die Spezifikation der Form der Vorhersagefunktion ist subjektiv, wenn beispielsweise der RGB-Scanner-Ausgang modelliert wird, können die Vorhersagefunktionen wie folgt gewählt werden: f1(dr, dg, db) = dr f2(dr, dg, db) = dg f3(dr, dg, db) = db f4(dr, dg, db) = dr 2 f5(dr, dg, db) = drdg f6(dr, dg, db) = drdb f7(dr, dg, db) = dg 2 f8(dr, dg, db) = dgdb f9(dr, dg, db) = db 2 To refine the forward model estimate, optional correction functions (as functions 525 in 5A shown). Since the actual physical process is not completely known, channel interactions can occur that affect the filter functions 510 and TRCs 520 do not reflect. The correction functions 525 represent these channel interactions. The correction functions are preferably multidimensional regression functions. The preferred regression correction functions may be formed as a linear combination of a set of prediction functions, which in turn are functions of the TRC output values. For the red channel, the regression function can be expressed as follows:

where d _r 'is the corrected model output value; d _r , d _g and d _{b are} the TRC output values; f _i (d _r , d _g , d _b ) is the ith prediction function; θ _{i is} the i-th regression _parameter and N _{pred is} the number of prediction functions in the regression. The specification of the shape of the predictive function is subjective, for example, when modeling the RGB scanner output, the predictive functions can be chosen as follows: f 1 (d r , d G , d b ) = d r f 2 (d r , d G , d b ) = d G f 3 (d r , d G , d b ) = d b f 4 (d r , d G , d b ) = d r 2 f 5 (d r , d G , d b ) = d r d G f 6 (d r , d G , d b ) = d r d b f 7 (d r , d G , d b ) = d G 2 f 8th (d r , d G , d b ) = d G d b f 9 (d r , d G , d b ) = d b 2

Given a given specification of the prediction functions, the regression parameters {β _i } can be determined via a standard standard regression process such as, for example, by S. Weisberg in "Applied Linear Regression," Chapter 8, pages 190 ff., John Wiley and Sons, New York, 1980 , is described. Stepwise regression with a confidence level of 95% can be done with digital values from the TRCs 520 and their combinations are used as regression variables and scanner RGB values from the scanner as data to be adjusted. The correction functions 325 often have a dominant linear term indicating that the forward model 500 takes most of the scanner's behaviors into account. Consequently, each correction function receives 525 all digital values from the TRCs 520 and each generates an estimated digital value for the R, G and B channels.

Photo printer model

at The photo printer model is an idealized printing device for determining spectral properties of the exposure illuminant over the Time. The photo printer model includes a printing lamp, filter, which can modify the printing illuminant, and an exposure table, which the respective duration and filter choice for a sequence of exposure periods specified. The printer model preferably comprises a color temperature component.

As in 9 is shown receives an exposure control 905 in the photo printer model 900 an exposure table supplied by the user and subtractive print filter transmission spectra. The exposure control 905 applies the exposure table to the subtractive print filter transmission spectra to obtain pressure filter transmission spectra over time. A parametric function 910 , preferably a blackbody radiator model, receives a lamp color temperature to model the print illuminant with a print illuminant power spectrum. Alternatively, the printing illuminant by a ge measured or user-specified spectrum. The computer system multiplies 915 the print filter transmission spectrum with the print luminance power spectrum to calculate the exposure illuminant spectrum over time.

As in 10 is shown, the computer system estimates in a process 1000 to estimate the exposure illuminant spectrum over time, the print illuminant spectrum (step 1005 ). As previously described, the print illuminant spectrum may be a supplied spectrum or a spectrum derived using a parametric model. The parametric blackbody radiator model can be expressed as follows: I (λT) = 3.74183 × 10 -16 λ (exp [0.014388 / λT] - 1) -1 where T is the color temperature in ° K, λ represents the wavelength and I is the light source spectrum thus generated. I (λ, T) is preferably normalized so that the spectral power at 560 nm is 1 or 100. From the calculated power spectrum, a main illuminant filter can be applied to reduce power in certain spectral regions, such as in the red or near-infrared region. This filter provides better emulation of the actual printing illuminants while still allowing for temperature variation using a single parameter. Commercial photo finishing equipment may include lamp house filters to adjust the color temperature of the printer beam to compensate for differences in paper sensitivities. To control the temperature of the printer beam, the color temperature of the blackbody radiator can be manually varied or a suitable measured illuminant spectrum can be used.

The computer system estimates the bulbs filter transmission spectra (step 1010 ). The filter transmission spectra are preferably measured by suitable physical photo printer filters. Illuminant filters in subtractive photoprinters may be dichroic filters having flat passbands and stop bands, as well as step transition regions, to remove specific wavelength ranges from the print illuminant. Additional printers can use red, green and blue bulbs. The computer system may preferably emulate additive printing with a subtractive system by using paired combinations of cyan, magenta, and yellow filters to achieve red, green, and blue exposures. Many techniques for modeling the exposure illuminant are possible, including parametric models and direct measurement of both additive and subtractive printer devices, filters and illuminants. For example, additive photo printers can use red, green and blue exposure illuminants. The chosen technique preferably specifies the spectral content of the exposure illuminant over time. The resulting function is preferably a function of time and wavelength depending on the user control.

The computer system applies the exposure filter transmission spectra to the print illuminant spectrum corresponding to the exposure table to estimate the exposure illuminant spectrum over time (step 1015 ). Consequently, the exposure illuminant spectrum has spectral characteristics that vary over time according to the exposure table. The exposure table specifies the lengths of time each print filter is switched into the main printer beam during exposure in the modeled printer. The exposure table may specify zero or more filters for a period of time. The exposure table serves as an interface for the user. The parameters of the exposure table correspond to those parameters that would be modified during a manual or automatic color correction.

An example 1100 an exposure table is in 11 shown. In this example 1100 For example, a subtractive printer is used in an additive printer mode with red, green and blue exposure periods. The spectral power of the printer illuminant is constant within each exposure period, and is the product of the printer illuminant multiplied by the product of the active filter transmission factors for the period.

Photo paper model

The Photographic paper model represented the behavior of real photo media both in terms of production the colors over the exposure and processing as well as in terms of colorimetric properties the color-based image thus generated. The paper model includes spectral sensitivities of the color layers, preferably CMY color layers. The paper model further includes parameters such as exposure color transfer properties, paper-based reflection and color absorption spectra of the paper. Each of these parameters is measured directly, estimated or estimated derived from measurements. The estimation process may be due the correlation between some of these paper parameters iteratively be.

As in 12 is shown multiplies the computer system in the photo paper model 1200 in step 1205 the film transmission spectrum of the scanner and film models, the exposure illuminant spectrum from the photo printer model and the spectral sensitivities of the paper. The computer system converts the spectral product into logarithmically integrated exposures 1210 , The computer system then converts the logarithmic integrated exposures to color concentration 1215 , The computer system guides the color concentrations to a printer media model 1220 to. The printer media model 1220 uses the paper reflection spectrum and the color absorption spectra to convert the input color concentrations into the predicted reflection spectrum for the modeled photo print.

typical Spectral sensitivity curves show the power needed to generate a specific density above the "haze" on photographic paper at any wavelength in the visible Spectrum is required. Photo paper manufacturers can give theirs Add paper to these curves. These added Curves can be taken into account by the power spectrum of the exposure illuminant or not. Spectral sensitivity curves through the spectral Bulk power have been normalized are referred to as "Equi power sensitivities". Equi-performance sensitivity curves can how so-called color matching functions are used and show the way photographic paper "perceives" an exposed negative image. The model is set assume that the sensitivities do not match the exposure level change, and fixed sensitivity curves produce acceptable predictions.

Sensitivity curves, based on iso-density contours are for spectral modeling the media usually not desirable. Densitometers usually do not provide a photocanal-independent measurement due to secondary absorptions the photo media colors. For example, a secondary absorption cause that the cyan hue to magenta and money coloring sensitivity values contributes. This property reduces the validity of the density based sensitivity curves. The color concentration creates however, a channel independent Measurement for photo media. Therefore, the photo paper model preferably uses Equi power sensitivity curves, derived from iso-concentration contours. This concentration-based Curves describe the power required at each wavelength is to produce a specific color concentration, and are a more faithful and desirable one representation the actual sensitivities the media.

wobei N_p die Anzahl von Belichtungsperioden ist, und e(λ, i)t_exp ¹ entsprechend das Belichtungsleuchtmittelspektrum und die Dauer für die i-te Belichtungsperiode sind. Die logarithmisch integrierte Belichtung H_x ist der Logarithmus der integrierten Belichtung und kann wie folgt ausgedrückt werden: Hx = log10Ex The color exposure characteristics of the photo media describe the amount of ink produced for a given channel as a function of the logarithmically integrated exposure for that channel. The integrated exposure E _x for a channel x can be expressed as follows: e x = ∫∫ λ T n (λ) e (λ, t) S x (Λ) dλdt where T _n (λ) is the spectral transmission factor of a color patch, e (λ, t) is the time-varying illumination illuminant, and S _x (λ) is the photo-paper sensitivity for channel x. If the spectrum of the exposure illuminant is constant within each period of a series of exposure periods (for example, a set specified by an exposure table), the integrated exposure can be expressed as follows:

where N _{p is} the number of exposure periods, and e (λ, i) t _exp ¹ corresponding to the exposure luminance spectrum and the duration for the ith exposure period. The logarithmically integrated exposure H _x is the logarithm of the integrated exposure and can be expressed as follows: H x = log 10 e x

Printouts the color concentration over the logarithmically integrated exposure for CMY channels are referred to as "C-logE" curves. C-logE curves preferably do not vary with a change in the Exposure illuminant as a direct consequence of the previously described Definition of the integrated exposure. Consequently, the C-logE curves for every exposure state can be used without recalculation.

C- log E curves are based on channel independent Measurements. Since C-logE concentration values between zero and the maximum concentration of each channel, the Concentration values by suitable adjustment of the color absorption spectra be normalized to 1.0. The color mixing model is also based on the color concentration as described above, and so-called C-logE curves allow a substantially direct calculation of the reflection of a pattern color.

The C-logE properties of processed photo paper depend on Development conditions, such as of the composition and Temperature of the developer solution and the development time. The photo paper model represents these Conditions relating to a single date. Unintentional changes These conditions mean that differences between actual Print and model-based forecasts exist.

intended Variations of development conditions can be made by adjusting the C-logE curves be modeled, either by replacing with alternatively measured Values or through parametric settings. A. E. Saunders describes such settings in "Fitting the Photographic Characteristic Curve: a Functional Equation Approach ", Journal of Photographic Science, Vol. 41, 1993, pages 186-93.

The spectral paper sensitivities and the C-logE curves describe the color generation process as a function of exposure. In particular, the C-logE curves convert the logarithmically integrated exposures into concentration values. For a channel x, the C-logE function can be expressed as follows: conc x = C x (H x ) where Cx is the C-logE function, represented for example in parametric or tabular form, H _{x is} the logarithmically integrated exposure described above, and conc _{x is} the resulting color concentration.

Photographic materials, such as paper and film, may show behavior known as reciprocity failure. The reciprocity error is a disturbance of the same trade-off between the time and the intensity of the exposure illuminant. This reciprocity can be seen in the calculation of the integrated exposure E _{x described} above. For very low values of the exposure time or the intensity of the illuminant, this compromise does not always apply exactly. This reciprocity error can be modeled by adjusting the inputs and / or outputs of the C-LogE calculation. Such modifications could be functions of logarithmically integrated exposure and exposure time. The functions could be implemented as parametric equations or lookup tables. For example, the C-LogE function for a channel x could be modified to include settings such as the following: conc x = C x [H rec.x (H x , t x , θ x )] where t _{x is} the exposure time and H _{rec.x is} a reciprocity error adjustment function with θ _x as the parameter _vector (or look-up table).

mit Rp(λ) = Rg(λ)exp[ – 2{cKC(λ) + mKM(λ) + γKy(λ)}]wobei λ die Wellenlänge repräsentiert, c, m, und y die mittels der C-logE-Kurven bestimmten Farbkonzentration sind, beispielsweise m = C_M(H_M)·K_C(λ), K_M(λ) und K_γ(λ) sind die normierten spektralen Farbabsorptionsvermögen des Fotopapiers, die mittels Faktoruntersuchung bestimmt werden, die derjenigen ähnelt, die zuvor in den Scanner- und Filmmodellen für die Filmfarbtöne beschrieben wurden. R_g(λ) ist das Reflexionsvermögen der Papierbasis, und R_p(λ) und R_p > (λ) sind entsprechend die nicht-korrigierten und korrigierten Spektral-Reflexionsvermögen eines Farb-Patches. Die in der Saunderson-Korrektur verwendeten Werte liegen im Bereich von k₁ = 0,04 und k₂ = 0,6, abhängig von dem Medium. Entsprechend ist R_p'(Σ) das vorhergesagte Reflexionsspektrum des modellierten Fotodruckes.The colorimetric properties of the sets of colors resulting from the C-logE calculations are preferably modeled using a Kubelka-Munk color mixing model and a Saunderson correction for surface refractive index differences. These colorimetric properties represent the predicted reflection spectrum of the modeled photo print and can be expressed as follows:

With R p (λ) = R G (λ) exp [- 2 {cK C (λ) + mK M (λ) + γKy (λ)}] where λ represents the wavelength, c, m, and y are the color concentration determined by means of the C-logE curves, for example m = C _M (H _M ) * K _C (λ), K _M (λ) and K _γ (λ ) are the normalized spectral color absorptivities of the photographic paper determined by factoring similar to those previously described in the scanner and film models for the film color tones. R _g (λ) is the reflectivity of the paper base, and R _p (λ) and R _p > (λ) are respectively the uncorrected and corrected spectral reflectances of a color patch. The values used in the Saunderson correction are in the range of k ₁ = 0.04 and k ₂ = 0.6, depending on the medium. Similarly, R _p '(Σ) is the predicted reflection spectrum of the modeled photo print.

Consequently does that Paper model the last calculation of the predicted reflection spectrum. However, as mentioned earlier, can do the calculation in a changing structure in another implementation carried out become. For example, the print model controls in an alternative Implementation the entire execution and determining the final predicted reflection spectrum. The print model requests the paper model for calculations, such as to the conversion of the logarithmically integrated exposures in concentrations and then around the reflection spectra for given Concentrations.

at One implementation will be the color absorption curves of the photographic paper in a similar way Way like those of the movie. Parts, the steps in each the color channels approaching the paper printed, and the prints thus produced and the paper base become measured with a spectrophotometer. The measured reflection spectra will be using the equations for the colorimetric properties, which were described in the photo paper model, in absorption spectra converted. As with the color negative film, creating is individual Color prints difficult to determine any absorption spectrum. Therefore, the factor analysis with these absorption spectra Tarot rotation applied to the desired absorption curves produce. As described above, eigenvalues of the absorption spectra of the Step wedge for each colorant can be used as the target spectra.

at an implementation will be used to determine the C-logE curves permeable Neutral step wedge using the modeled photo printing process printed. A neutral step key includes substantially gray patterns in discrete steps from low density to high density. The reflection spectra the printed step wedge pattern is measured, as well as the transmission spectra the neutral step wedge pattern. Logarithmically integrated exposures for each transmissive wedge patterns are determined as previously described. The Color absorbance of the paper are detected and the color concentrations become for each printed patterns using a color formulation process determined, for example, by the process described by R. S. Berns in "Spectral modeling of a dye diffusion thermal transfer printer, Journal of Electronic Imaging, Vol 2, No. 4, October 1993, pages 359-70. For each Hue will be the paired concentration and the logarithmic integrated exposure values entered into a spline function. The resulting splines will be at a uniform distance again sampled.

at an implementation will be used to derive the spectral sensitivities of photographic paper experimentally determined curves or estimated curves for initializing parameters of a search process for the previously described so-called best-fit sensitivities used. There the C-logE curves from the spectral sensitivities for the calculation of the logarithmic depend on integrated exposure, The C-logE curves are updated as the sensitivities change. An iterative process for setting the sensitivities and subsequently the C-logE curves are performed, until a stop criterion is fulfilled is, for example, if the curves converge sufficiently.

implementation

The Invention may be in hardware, software or in a combination of Hardware and software are implemented. However, the invention is preferred implemented in computer programs running on programmable computers run, such computers, a processor, a data storage system (with fleeting and non-volatile Memory and / or memory elements), at least one input device and at least one output device. To enter data a program code is used ver to the previously described function perform and to generate output information. The initial information be in a known manner one or more output devices fed.

each This program may be written in any desired computer language (including machine language, Assembly language, higher procedural programming language or object-oriented Programming language) to be implemented with a computer system to communicate. The language may be a compiler language or to act an interpreted language.

One such computer program is preferably stored on a storage medium or a storage device (for example, ROM, CDROM or magnetic Media), wherein the storage medium or the storage device from a general purpose computer or from a special purpose programmable Calculator can be read to configure the computer and to use when reading the storage medium from the computer is to carry out the methods described above.

It has become a number of implementations of the present invention described. It should be noted, however, that different Modifications possible are without departing from the scope of the present invention. Other variations are possible such as the use of CMY color space coordinates for the forward model instead of RGB color space coordinates. Alternatively you can the models for the components (for example the negative, the colors, the paper, etc.) are selected from a database of existing models. The filter functions can also be provided or measured directly.

Claims (24)

A method of modeling spectral characteristics of a photographic print on a photographic negative comprising the steps of: converting a digital value obtained by scanning a photographic negative into a film pass spectrum ( 205 ) using a photographic negative film model ( 105 ) corresponding to a photographic negative medium; - estimating a recording lighting spectrum ( 210 ) over time using an exposure schedule specifying the periods of application of a print illuminant over at least one exposure period; - estimating paper spectral sensitivities ( 215 ) of photographic paper corresponding to the photographic printout; Integrating Spectral Products of the Film Transmission Spectrum, the Recording Illumination Spectrum and the Paper Spectral Sensitivities ( 220 ) over time; - Converting the integrated spectral products to log-integrated exposures ( 225 ); - Convert the log-integrated exposures into Dye concentrations ( 230 ); and converting the dye concentrations into a predicted reflection spectrum ( 235 ). The method of claim 1, wherein converting digital values to a movie pass spectrum comprises the steps of: iteratively updating media coordinates according to an estimated spectrum until a stop criterion ( 320 ), wherein the media coordinates are in the colorant space of a pattern scanned by an image capture device; and restricting the updating by means of a media model ( 310 ) according to the pattern using a limited minimization process. The method of claim 1, wherein converting digital values to a movie transmission spectrum comprises the steps of: (a) estimating media coordinates ( 405 ) in a colorant space of the photographic negative scanned by an image capture device, the estimated media coordinates corresponding to target digital values generated by the image capture device when scanning the photographic negative; (b) converting the estimated media coordinates into an estimated spectrum ( 410 ) using a photographic negative film model corresponding to the photographic negative; (c) estimating digital values by providing the estimated spectrum to a forward model that comprises the image capture device (10); 415 ) modeled; (d) identifying a film error between the estimated digital values and the target digital values ( 420 ); and (e) if a stop criterion is not met ( 425 ), Scanning the colorant space for media coordinates according to the film error, and repeating steps (b) through (e) until the stop criterion is met ( 430 ). The method of claim 3, wherein: - the media coordinates for the photographic negative film model dye concentrations of cyan, magenta and and yellow dyes; and - the Media co-ordinates containing the film error within the film tolerance generate the film transmission spectrum. The method of claim 3, wherein the searching of the colorant space includes using a numerical minimization technique that is so limited is that no media coordinates out of bounds accordingly to be identified with the photographic negative. The method of claim 3, wherein the image capture device comprises at least one output channel and the forward model for each output channel of the image capture device comprises the steps of: - multiplying ( 505 ) of a pattern spectrum transmitted to the forward model ( 305 ) and a value of a filter function ( 510 ) for generating a spectral product, the value of the filter function representing a product of a filter transmittance spectrum corresponding to a filter of the image sensing device and a spectrum of illuminants corresponding to a light source; - Integrate ( 515 ) of the spectral product over at least one wavelength to produce an integral value; and converting the integral value to an estimated digital value by applying a one-dimensional transformation function ( 520 ) to the integral value. A method according to claim 6, wherein the spectral product is over the visible wavelengths is integrated. The method of claim 6, further comprising the step of applying a multi-dimensional regression function ( 525 ) on the estimated digital values for modeling a channel interaction in the image capture device. Method according to Claim 6, in which the one-dimensional transformation function ( 520 ) is a gain offset gamma function, the method further comprising estimating parameters of the gain offset gamma function ( 610 ) and such estimating comprises the following steps: - scanning a flat spectra target to generate flat pattern values ( 605 with the flat spectra target having a neutral color wedge such that the spectra are substantially constant at each visible wavelength; - Measurement of flat pattern spectra ( 607 ) of the flat spectra target ( 607 ); and estimating gain, offset and gamma parameters ( 610 ) using flat pattern values and flat pattern spectra using a minimized minimization technique. The method of claim 6, further comprising the step of estimating filter parameters of the filter function ( 625 ), such estimation comprising the steps of: - scanning a multicolor target ( 615 ) for generating multicolor pattern values, the multicolor target including colors that span a scale of a printer device that produced the multicolor target; - measuring multicolor pattern spectra ( 620 ) of the multicolor target; - estimating filter parameters ( 625 ) the filter radio based on the multicolor pattern values and the multicolor pattern spectra using a minimized minimization technique; - calculating filter digital values ( 630 by transmitting the multicolor pattern spectra to the forward model and applying the estimated filter functions; - Calculate a filter difference ( 635 by comparing the filter digital values with the multicolor pattern values; and - if a filter stop criterion ( 640 ) is not met, re-estimating the filter parameters ( 650 ) based on the filter difference and repeating the steps of calculating the filter digital values, calculating the filter difference, and re-estimating the filter parameters until the filter stop criterion is met. The method of claim 10, wherein the one-dimensional transform function is a gain offset gamma function, and the method further comprises the steps of: estimating gain offset gamma parameters ( 805 ) for the gain offset gamma function; Calculating design digital values using the gain offset gamma function and the filter function with the respective estimated parameters ( 815 ); - calculating a design difference ( 820 ) based on the design digital values and the multicolor pattern values; and - if a convergence criterion is not met, re-estimating the gain-offset gamma parameters and the filter parameters ( 830 . 835 ) until the convergence criterion is met. The method of claim 11, wherein estimating the gain offset gamma parameters comprises the steps of: - scanning a flat spectra target ( 605 ) for generating flat pattern values, the flat spectrum target including a neutral color wedge such that the spectrums are substantially constant at each visible wavelength; Measurement of flat pattern spectra ( 607 ) of the flat-spectrum target; and estimating gain, offset and gamma parameters ( 610 ) using flat pattern values and flat pattern spectra using a minimized minimization technique. The method of claim 6, wherein the one-dimensional transformation function is a gain-offset gamma function and the method further comprises the steps of: - scanning a flat-spectrum target ( 605 ) for generating flat pattern values, the flat spectrum target having a neutral color wedge such that the spectrums are substantially constant at each visible wavelength; - Measurement of flat pattern spectra ( 607 ) of the flat-spectrum target; - scanning a multicolor target ( 615 ) for generating multicolor pattern values, the multicolor target including colors that span a scale of a printer device that produced the multicolor target; - measuring multicolor pattern spectra ( 620 ) of the multicolor target; - determining flat mean values based on the flat pattern values; Estimation of gain, offset and gamma parameters ( 610 ) for the gain offset gamma function from the flat mean values and the flat pattern spectra using a limited minimization; - estimating filter parameters ( 625 ) the filter functions based on the multicolor pattern values and the multicolor pattern spectra using limited minimization; - re-estimating the gain offset gamma parameters ( 830 ) on the basis of the flat pattern spectra; - re-estimating the filter parameters ( 835 ) based on the newly estimated gain offset gamma functions; - calculating design digital values ( 815 using the gain offset gamma function and the filter function with the respective re-estimated parameters from the multicolor pattern spectra; - calculating a design difference ( 820 ) based on the design digital values and the multicolor pattern values; and, if a convergence criterion is not met, re-estimating the gain offset gamma parameters and the filter parameters ( 830 . 835 ) until the convergence criterion is met. The method of claim 6, wherein the image capture device has at least one output channel and the forward model for each output channel of the image capture device comprises the steps of: - multiplying ( 552 ) a sample spectrum transmitted to the forward model and a light source spectrum corresponding to a light source for generating a first spectral product; - Multiply ( 555 ) of the first spectral product with a filter transmission spectrum ( 560 ) corresponding to a filter of the image capture device for generating a second spectral product; - Integrate ( 565 ) of the second spectral product over at least one wavelength to produce an integral value; and converting the integral value to an estimated digital value by applying a one-dimensional transformation function ( 570 ) to the integral value. Method according to one of claims 1 to 4, wherein estimating an exposure illumination spectrum comprises the steps of: estimating a pressure illumination spectrum ( 1005 ); Estimating Bulb Filter Permeability Spectra Using the Pressure Illuminant ( 1010 ) modify; Applying the estimated luminous flux filter permeability spectra to the pressure illuminant spectrum ( 1015 ) according to the exposure plan, the exposure plan further specifying the periods and filter selections for the at least one exposure period. The method of claim 15, wherein the exposure plan over at least specifies a period of time in which at least one illuminant filter exposed to the pressure bulb. The method of claim 15, wherein the pressure-illuminant spectrum is a measured spectrum of a light source. The method of claim 15, wherein the pressure-illuminant spectrum derived either from a parameter model or a blackbody model is. The method of claim 15, wherein the illuminant filter transmission spectra model is either as a filter for a photographic subtractive printer device or for a photographic Additive printer apparatus acts. A method according to any one of claims 1 to 19, wherein the paper spectral sensitivities of Isoconcentration contours derived energy equal sensitivity curves are and one at each wavelength for producing a specified dye concentration in the Photograph paper required energy. A method according to any one of claims 1 to 19, wherein the paper spectral sensitivities are dye exposure transfer characteristics include a lot of for a given channel depending dye produced by the log-integrated exposure for this channel represent. A method according to any one of claims 1 to 21, further comprising Step of estimating of dye absorption curves for the photo paper, being the treasure the dye absorption curves comprises the following steps: - Measure up a paper base spectrum of the photographic paper; - Measure up a paper-target spectrum of a paper target with steps in each of the colorant channels the photo paper; - Convert the paper base spectrum and the paper-target spectrum in absorption spectra; and - Apply a factor analysis and a target rotation on the absorption spectra for generating dye absorption curves. A method according to any one of claims 1 to 22, further comprising Step of determining C-logE curves, the graphs of dye concentration over log-integrated exposure for colorant channels are, wherein determining the C-logE curves comprises the steps of: - To Print of a neutral step wedge with gray patterns in essentially discrete Low to high density steps on the photo paper; - Measure up curve reflectance spectra of the printed neutral step wedge patterns; - Measure up of curve transmission spectra the pattern of the printed neutral step wedge; - Determine a log-integrated exposure for each pattern of the printed Neutral step wedge; - Determine dye absorption degrees of the photographic paper; - Determine of dye concentrations for each pattern of the printed neutral step wedge using the particular dye absorption levels; - Apply a template matching function on paired dye concentrations and log-integrated exposures for creating templates; and - resampling the templates with even spacing to generate the C-logE curves. Computer program in a computer readable medium for modeling spectral characteristics of a photographic Expression based on a photographic negative, the computer program Has commands that cause a computer to perform the procedure according to claims 1-23 performs.